Full Resolution Image Compression with Recurrent Neural Networks
Venue
arxiv (2016)
Publication Year
2016
Authors
George Toderici, Damien Vincent, Nick Johnston, Sung Jin Hwang, David Minnen, Joel Shor, Michele Covell
BibTeX
Abstract
This paper presents a set of full-resolution lossy image compression methods based
on neural networks. Each of the architectures we describe can provide variable
compression rates during deployment without requiring retraining of the network:
each network need only be trained once. All of our architectures consist of a
recurrent neural network (RNN)-based encoder and decoder, a binarizer, and a neural
network for entropy coding. We compare RNN types (LSTM, associative LSTM) and
introduce a new hybrid of GRU and ResNet. We also study "one-shot" versus additive
reconstruction architectures and introduce a new scaled-additive framework. We
compare to previous work, showing improvements of 4.3%-8.8% AUC (area under the
rate-distortion curve), depending on the perceptual metric used. As far as we know,
this is the first neural network architecture that is able to outperform JPEG at
image compression across most bitrates on the rate-distortion curve on the Kodak
dataset images, with and without the aid of entropy coding.
